Robust edge detection using Mumford-Shah model and binary level set method

Abstract

A new approximation of the Mumford-Shah model is proposed for edge detection, which could handle open-ended curves and closed curves as well. The essential idea is to treat the curves by narrow regions, and use a sharp interface technique to solve the approximate Mumford-Shah model. A fast algorithm based on the augmented Lagrangian method is developed. Numerical results show that the proposed model and method are very efficient and have the potential to be used for edge detections for real complicated images. © 2012 Springer-Verlag.